Confluence export: every way out, and what each one keeps

TL;DR. Confluence has four ways out: single-page export (PDF or Word), space export (HTML, XML, or PDF), the full-site Atlassian Cloud export, and the REST API. PDF and Word keep how a page looks and lose how it is structured. HTML and XML space exports keep the page tree and attachments and lose most live macros. The API keeps everything but asks you to write code. If your goal is a new wiki rather than an archive, the move is to reach a portable Markdown format and import that.

A Confluence export is how you get your content out of Confluence — and the right method depends on whether you want a document to print, an archive to keep, or a wiki to move into. Confluence ships several export paths, each preserving a different slice of what a page actually is. Pick the wrong one and you end up with a beautiful PDF of a wiki you can no longer edit. (If you’ve ever opened a 400-page space PDF looking for one runbook, you have suffered enough.)

This post walks every export path Confluence offers, what each one keeps and what it quietly drops, and how to land on a format a different wiki can read. We’re the company that wants you to import the result, so squint at the recommendation at the end — but the mechanics in the middle are the same no matter where you’re going.

Confluence has four doors, not one

People say “export Confluence” as if it’s a single button. It isn’t. There are four distinct mechanisms, and they sit at different levels of your workspace.

Single-page export. From any page’s more-actions menu, you can export that one page to PDF or Word. This is the door most people find first, because it’s the one in front of them on the page they’re reading.
Space export. From Space Settings, a space admin can export an entire space to HTML, XML, or PDF. This is the door you want for migration — it carries the page tree and the attachments.
Full-site Atlassian Cloud export. An organization admin can export the whole site. This is a backup-and-restore mechanism, not a per-team tool, and it produces a Confluence-native bundle meant to be restored into Confluence.
The REST API. Confluence exposes page content programmatically. If you can write a script, you can pull every page as storage-format XHTML and do whatever you like with it.

Most teams only ever need doors one and two. The decision between them is almost always “am I keeping a document, or am I moving a wiki?” Keep that question in your head for the rest of the post.

Single-page export keeps the look and loses the structure

The single-page export is the fastest door and the most misleading. You open a page, click the three dots, choose Export to PDF or Export to Word, and you have a file in seconds. The file looks right. That’s the trap.

PDF is a faithful snapshot. Layout, images, headings, and formatting all survive. What doesn’t survive is everything that made it a wiki page: it’s now read-only, the internal links point nowhere useful, the macros are frozen as whatever they happened to render, and the page tree it lived in is gone. PDF is the right answer when you need to send a page to someone outside the wiki — a contract, a one-pager, an audit artifact. It is the wrong answer when you want to move the page.

Word gives you an editable .docx. That sounds better than PDF and is usually worse for migration. The conversion mangles tables, flattens macros, and produces formatting that looks fine until you scroll to the table someone built with eleven merged cells. Word export is for the colleague in Legal who wants to redline a policy, not for the engineer rebuilding a knowledge base somewhere else.

Neither single-page path gives you structured, re-importable content. Both are document exports. A wiki page isn’t a document — it’s a node in a tree, with links to other nodes, and the export that forgets the tree forgets most of the point.

Space export keeps the tree, the attachments, and most of your sanity

This is the door that matters for migration. From Space Settings → Content Tools (or Export Space, depending on your Confluence version), a space admin can export the entire space. Three formats:

HTML export produces a static, browsable copy of the space — rendered pages, a navigation index, and the attachments in a folder. You can open it in a browser with no Confluence running. It’s an honest archive, and it’s the format most third-party wiki importers read, because rendered HTML is a known quantity. The page tree survives as a folder hierarchy. Attachments come along.

XML export is the Confluence-native one. It contains the storage-format representation of every page, plus page history and metadata, bundled for restoring into another Confluence instance. It is not meant for humans to read, and opening one in a text editor is a fast way to develop opinions about angle brackets. But it carries the most fidelity, which is why migration tools that target Confluence-to-X conversions often read it.

PDF space export flattens the whole space into one long document. It’s the archive equivalent of printing your house. Use it when someone in compliance needs “the whole space as of today” in one file and never wants to click anything.

Atlassian documents the official mechanics — formats, permissions, the background-job behaviour on large spaces — on their support site. The canonical reference is Atlassian’s export documentation, and it’s worth a read before you commit to a path, because the available formats differ slightly between Cloud and Data Center.

A table for the door you’re standing in front of

Here’s the whole decision in one place. Find your export method, read across, and you’ll know what you’re keeping before you click.

Export method	What it preserves	What it loses
Single-page PDF	Layout, images, formatting, inline content	Editability, internal links, page tree, live macros
Single-page Word	Editable text, headings, basic formatting	Tables (often), macros, attachments reliability, page tree
Space HTML	Rendered pages, page tree as folders, attachments	Live macros (frozen to static), page history, restrictions
Space XML	Storage-format pages, page history, metadata	Human readability, easy reuse outside Confluence
Space PDF	Whole space as one document	Structure, links, editability — everything but the words
Full-site Cloud export	Entire site, native fidelity, for restore	Per-team usability — it’s an admin backup, not a doc
REST API	Everything, page by page, as XHTML	Your evening, unless you like writing scripts

The pattern across the table: the easier the export is to click, the less of your wiki it keeps. PDF is one click and keeps the least. The API is the most work and keeps the most. HTML and XML sit in the middle, which is exactly why they’re the migration formats.

The full-site export and the API are for different jobs

The full-site Atlassian Cloud export is an organization-admin function. It produces a complete backup of every space, meant to be restored into another Confluence instance or kept for disaster recovery. It is not a tool for “my team is leaving for a different wiki” — reach for it to migrate one team’s docs and you’ll get a bundle shaped for Confluence’s restore mechanism and nothing else.

The REST API is the power-user door. Every page is available as storage-format content through the content endpoints; you can walk the page tree, pull each body, fetch attachments, and write the result wherever you want. It’s the most faithful and most flexible path, and the one that turns “export my wiki” into “write and maintain an export script.” For a one-time migration it’s overkill — the space HTML export plus a good importer gets you there without a single line of code. The API exists for the same reason this post does: Confluence’s content is yours, and an honest export doesn’t hide behind a sales call.

Markdown on demand is the right shape. Talk-to-sales is the wrong one.

This is the opinion, and it’s a hill we’ll defend with a number.

Every export path above is, fundamentally, Confluence answering the question “can I have my data back?” Some products answer that question with a self-serve button. Others answer it with a contact form and a calendar invite. Markdown on demand is the right shape; talk-to-sales is the wrong one. A wiki that makes you schedule a call to get your own content out has already told you how it plans to treat you.

Raccoon Page exports a per-space zip of Obsidian-compatible Markdown and per-page Markdown with one click, on every plan, including the Free tier — $0, three users, one space, one hundred pages, no card. There’s no export tier, no “enterprise data egress” line item, no call. The per-space zip export is right there in Space Settings, and what comes out is a Markdown vault you could read in a text editor or drop into any tool that reads Markdown. We built it that way on purpose. You’re never locked in, which is easy to say and harder to ship — so we shipped it, and then we wrote it down.

The contrast we keep in-house has a name. In Raccoon Corp lore, Bandito von Scavengix has seen things in Confluence export files — specifically the raccoon-to-rabbit emoji incident, where an inherited “small-mammal normalisation” lookup table in an export helper quietly turned every raccoon emoji into a rabbit. PR #2 was the fix: delete the table, because we’re an employer that includes raccoons. The lesson generalises. An export that silently changes your content into a different animal isn’t an export. It’s a betrayal with a progress bar.

When you should not export at all

Honest section, because the export you don’t run is sometimes the right call.

If your team is heavily on Jira and the Jira-and-Confluence loop is doing real work, exporting your wiki to leave is solving the wrong problem. The integration is the moat, and the speed pain you’re feeling may not be worth unwinding twenty linked spaces. Stay, and revisit in a year.

If you only need to share one page with someone outside the wiki, don’t export the space — export the single page to PDF and email it. Running a full migration to hand a contractor one runbook is the export equivalent of moving house to change a lightbulb.

And if you’re exporting because search is bad and pages are slow, be honest about whether a new tool fixes the cause or just moves the symptom somewhere prettier. The migration is real work. Run it when the destination is decided, not as a way of avoiding the decision. Our Confluence alternatives roundup is the read for where to land before you pull anything out.

Reaching a format a new wiki can read

If you’ve decided to move, the path is short. You don’t need PDF, you don’t need Word, and you almost certainly don’t need the API. You need the space HTML or XML export and an importer that reads it.

Run the space export from Space Settings. Download the zip. Then hand it to the destination wiki’s importer, which does the conversion you’d otherwise be scripting: rendered HTML or storage XML in, structured Markdown-compatible content out, page tree and attachments preserved. The Confluence import into Raccoon Page reads the export, keeps the folder hierarchy and images, and converts the most common macros — info, tip, note, and warning callouts, code blocks, status macros — on a best-effort basis. Most wikis import in under 10 minutes. Sub-second loads. Keyboard-first. That’s the wiki the export is walking toward.

The export is the easy half. The hard half is the team agreement — deciding to leave, picking where to go, getting everyone to write in the new place. No export format solves that. But once the agreement is made, the mechanics are an afternoon, not a quarter.

Things people actually ask

How do I export a single Confluence page? Open the page, use the more-actions menu (the three dots near the top right), and choose Export. Confluence offers PDF and Word for a single page. PDF preserves layout and is read-only; Word gives you an editable .docx but mangles macros and tables. Neither produces structured content you can re-import cleanly into another wiki.

How do I export a whole Confluence space? Go to Space Settings, then Content Tools or Export Space depending on your version. You can export a space to HTML, XML, or PDF. HTML gives you a browsable static copy with attachments. XML is the Confluence-native backup format used for restore and migration. PDF flattens the whole space into one document.

What is the difference between Confluence HTML and XML export? HTML export produces a static, browsable copy of the space with rendered pages and attachments — good for archiving and for importers that read rendered HTML. XML export produces the Confluence-native storage format including page history and metadata, intended for restoring into another Confluence instance rather than for human reading.

Can I export Confluence to Markdown? Not directly from Confluence itself. Confluence has no native Markdown export. You either run the HTML or XML space export and convert it, or use a wiki importer that reads the Confluence export and writes Markdown on the way in. The Raccoon Page Confluence importer does exactly that.

Does Confluence export keep attachments and images? Space-level HTML and XML exports include attachments and images in the bundle. Single-page PDF embeds images inline. Single-page Word is the least reliable for attachments. The full-site Atlassian Cloud export includes everything but is an admin-only backup, not a per-team tool.

How long does it take to export a Confluence space? A single page exports in seconds. A space of a few hundred pages exports in a minute or two. Very large spaces or a full-site export can take much longer and may run as a background job. Importing the result into Raccoon Page takes under 10 minutes for most wikis.

What does Confluence export lose? Macros are the main casualty. Dynamic macros that query Jira or render live data export as static snapshots or empty placeholders. Page restrictions, version history beyond the latest revision, and some inline metadata are not preserved by every path. PDF and Word also discard the page tree.

Do I need admin rights to export Confluence? Single-page PDF and Word exports are available to anyone who can view the page. Space export usually requires space-admin permission. The full-site Atlassian Cloud export requires organization-admin rights and is meant for backup and migration, not everyday use.

Confluence has more ways out than people think, and most of them keep less than people hope. PDF for the page you’re mailing, HTML or XML for the wiki you’re moving, the API for the script you’d rather not write. Pick the door by the question — document, or wiki? — and the rest is an afternoon. If the answer is wiki, the ten-minute Confluence import is the short walk from the export you just ran to a wiki that hands your content back the same way it took it in: as Markdown, on demand, no call required. Bandito will check it for rabbits.